An early warning system for flooding risk prevention in estuaries

Flooding events constitute a serious risk to human safety and infrastructure in coastal communities. In estuarine environments, extreme sea level events caused by the combined effects of mean sea level anomalies, astronomical tides and storm surge pose significant threats to urban areas as well as critical assets inside the estuary. 

To address these challenges, this work presents an Early Warning System for flood risk prevention in estuaries, developed within the COSNORTH project under the framework of the Copernicus National Marine Service Collaboration Programme: EU Coastal Monitoring Demonstrators.

The system is based on the following components:

• An Operational Estuarine Forecast Modelling System nested within the Copernicus Marine Service to provide high-resolution hydrodynamic variables (~50 m spatial resolution) in Santander Bay. A twofold approach has been applied in the estuary, combining deep learning techniques (RNN-LSTM) with dynamic downscaling results based on Delft3D.
• Uncertainty estimates to define confidence intervals for extreme sea level forecasts, derived from historical records by quantifying and sampling error distributions from comparisons between simulated and observed tide and non-tidal residuals during representative extreme events.

• Flooding risk warning levels, categorized into three levels: none, medium, and high. The warning levels are calculated taking into account the coastal and infrastructure flood exposure and the sea level forecasts and confidence intervals previously mentioned.
• A Web App, which allows users to interact with flood risk data, offering detailed insights into affected areas and potential impacts. The platform also allows users to interactively report flooding events, building an observational database that will be used to validate and continuously improve the forecasting system.

As a result, the system provides daily 3-day forecasts of total sea level, astronomical tide and storm surge, as well as the warning levels for floodrisk. The system has been validated using tide gauge data from Puertos del Estado (Spain), showing a high accuracy in modelling storm surge, astronomical tide and total sea level. A comprehensive description of the system implementation and validation will be provided in the presentation, including an intercomparison of sea level simulations derived from AI-based approaches and numerical hydrodynamic modelling, as well as a quantitative uncertainty assessment.